Re: [WIRELESS-LAN] WLAN Operating costs

[John J. Brassil]

> I'm trying to get a sense of the costs associated with operating a WLAN,
> specifically as it relates to "physical" maintenance of a network with
> 100s of APs.
>
> In other words:
>
> * How often does one need to "touch" an AP for maintenance purposes
> (upgrade, replacement, configuration, debug, reset, etc.) ?
>
> * How much time does one typically "spend" with the AP in the above
> cases and how is that time impacted by the fact the AP is in a
> hard-to-reach place (e.g. ceilings) ?
>
>
>
> Can anyone share their experience and/or statistics ?
>
>
>
>
>
> Yair

Two words: Inline power. 95+% of your problems can be solved by a reboot and toggling the switch port on & off has that effect, at least on Cisco gear. Mmmm, no more trips up the ladder...


In related news, *any* config change to the port also has this effect, so keep that in mind. After we installed our NMS from Airwave, I set it to email all our engineers on three conditions: Rogue, Config Mismatch and Device Down. We started getting lots of alerts for devices that we weren't getting paged on. The reason: Our main NMS (Spectrum) has a 2 minute delay built in to avoid annoying the crap out of us with transient alerts(usually polling misses) so we weren't seeing the vast majority of these. After a bit of digging and some great guesswork, we noticed that most of these alerts were coming from devices on POE ports & while we weren't able to *conclusively* prove anything, at least one possible cause appeared to be auto-negotiation issues. Now, these are Cisco devices on both ends, so you would think they would figure out 100/full and leave it like that but as anyone who has operated a network of any appreciable size for very long knows, the mysteries of autonegotiation are deep and dark indeed. So since we also encountered some other discrepancies in our AP port configs (OK, quickly, what's the difference between switchport mode dynamic and switchport mode dynamic desrireable?) we decided to write a script to standardize them all, including explicit speed/duplex settings for 100/full.

Net result: AP reboots on POE switchports have dropped to nearly none. Not quite none, but we're now getting like 1 a week vs. 6-10 a day (out of about 300 APs using POE.) We still need to figure out what's up with those remaining devices but it appears we've done a pretty good job of improving uptime in our network. Which is, after all, why we get paid the big bucks. ;)

So, the moral of this story is: Don't trust Cisco. :) No, really, I guess it would be be even something really great (POE) can have have unanticipated drawbacks. Hopefully in the future Cisco (and other vendors?) will decouple config changes from power delivery, but for now, hopefully this will save some of you the hassle we went through.

Have a great weekend everybody,

John

John